Skip to content

data: remove 4 test smoke sequences that are not wildfires - #47

Closed
MateoLostanlen wants to merge 2 commits into
mainfrom
data/remove-non-fire-test-smoke
Closed

MateoLostanlen wants to merge 2 commits into
mainfrom
data/remove-non-fire-test-smoke

Conversation

@MateoLostanlen

Copy link
Copy Markdown
Member
  • A viewer review of the v4.5.0 test (temporal-model) found 4 test smoke sequences that are not really wildfires. They are removed completely: registry entries and raw folders, so they leave both the sequential and the YOLO datasets.
    • awf-axis_elephant_999_2023-06-03T20-44-26
    • pyronear-sdis-77_croix-augas_059_2024-09-01T20-56-15
    • pyronear-sdis-77_croix-augas_059_2024-09-03T14-52-45
    • sis-67_col-saverne-02_5_2026-07-25T13-15-01 (annotator import)
  • The annotator one keeps its import_plan.json entry with an "excluded" reason. Planning keeps existing entries, so it is not picked again, and materialisation now skips excluded entries, so add_data never sees it.
  • sequential_test goes from 223 to 219 WF (877 FP unchanged). This shrinks test, which the release notes must say. CLAUDE.md documents how to remove a sequence.

Draft: the data commit (registry pointer, dvc.lock) lands after the rebuild and leakage checks. More review decisions may be added before release.

A sequence removed from the dataset after review keeps its import_plan
entry with an 'excluded' reason, so a later planning run does not pick
it again, and materialisation no longer stages it for add_data.
Found in a viewer review of the v4.5.0 test. Registry entries and raw
folders are deleted; the annotator-sourced one is flagged excluded in
import_plan.json so it is not re-imported. sequential_test is now
219 WF + 877 FP.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant